Train Small, Infer Large: Memory-Efficient LoRA Training for Large Language Models

Large Language Models (LLMs) have significantly advanced natural language processing with exceptional task generalization capabilities. Low-Rank Adaption (LoRA) offers a cost-effective fine-tuning solution, freezing the original model parameters and training only lightweight, low-rank adapter matrices. However, the memory footprint of LoRA is largely dominated by the original model parameters. To mitigate this, we propose LoRAM, a memory-efficient LoRA training scheme founded on the intuition that many neurons in over-parameterized LLMs have low training utility but are essential for inference. LoRAM presents a unique twist: it trains on a pruned (small) model to obtain pruned low-rank matrices, which are then recovered and utilized with the original (large) model for inference. Additionally, minimal-cost continual pre-training, performed by the model publishers in advance, aligns the knowledge discrepancy between pruned and original models. Our extensive experiments demonstrate the efficacy of LoRAM across various pruning strategies and downstream tasks. For a model with 70 billion parameters, LoRAM enables training on a GPU with only 20G HBM, replacing an A100-80G GPU for LoRA training and 15 GPUs for full fine-tuning. Specifically, QLoRAM implemented by structured pruning combined with 4-bit quantization, for LLaMA-3.1-70B (LLaMA-2-70B), reduces the parameter storage cost that dominates the memory usage in low-rank matrix training by 15.81$\times$ (16.95$\times$), while achieving dominant performance gains over both the original LLaMA-3.1-70B (LLaMA-2-70B) and LoRA-trained LLaMA-3.1-8B (LLaMA-2-13B).

Preventing the Popular Item Embedding Based Attack in Federated Recommendations

Privacy concerns have led to the rise of federated recommender systems (FRS), which can create personalized models across distributed clients. However, FRS is vulnerable to poisoning attacks, where malicious users manipulate gradients to promote their target items intentionally. Existing attacks against FRS have limitations, as they depend on specific models and prior knowledge, restricting their real-world applicability. In our exploration of practical FRS vulnerabilities, we devise a model-agnostic and prior-knowledge-free attack, named PIECK (Popular Item Embedding based Attack). The core module of PIECK is popular item mining, which leverages embedding changes during FRS training to effectively identify the popular items. Built upon the core module, PIECK branches into two diverse solutions: The PIECKIPE solution employs an item popularity enhancement module, which aligns the embeddings of targeted items with the mined popular items to increase item exposure. The PIECKUEA further enhances the robustness of the attack by using a user embedding approximation module, which approximates private user embeddings using mined popular items. Upon identifying PIECK, we evaluate existing federated defense methods and find them ineffective against PIECK, as poisonous gradients inevitably overwhelm the cold target items. We then propose a novel defense method by introducing two regularization terms during user training, which constrain item popularity enhancement and user embedding approximation while preserving FRS performance. We evaluate PIECK and its defense across two base models, three real datasets, four top-tier attacks, and six general defense methods, affirming the efficacy of both PIECK and its defense.

In-Dataset Trajectory Return Regularization for Offline Preference-based Reinforcement Learning

Offline preference-based reinforcement learning (PbRL) typically operates in two phases: first, use human preferences to learn a reward model and annotate rewards for a reward-free offline dataset; second, learn a policy by optimizing the learned reward via offline RL. However, accurately modeling step-wise rewards from trajectory-level preference feedback presents inherent challenges. The reward bias introduced, particularly the overestimation of predicted rewards, leads to optimistic trajectory stitching, which undermines the pessimism mechanism critical to the offline RL phase. To address this challenge, we propose In-Dataset Trajectory Return Regularization (DTR) for offline PbRL, which leverages conditional sequence modeling to mitigate the risk of learning inaccurate trajectory stitching under reward bias. Specifically, DTR employs Decision Transformer and TD-Learning to strike a balance between maintaining fidelity to the behavior policy with high in-dataset trajectory returns and selecting optimal actions based on high reward labels. Additionally, we introduce an ensemble normalization technique that effectively integrates multiple reward models, balancing the tradeoff between reward differentiation and accuracy. Empirical evaluations on various benchmarks demonstrate the superiority of DTR over other state-of-the-art baselines

A Comprehensive Study of Shapley Value in Data Analytics

Over the recent years, Shapley value (SV), a solution concept from cooperative game theory, has found numerous applications in data analytics (DA). This paper provides the first comprehensive study of SV used throughout the DA workflow, which involves three main steps: data fabric, data exploration, and result reporting. We summarize existing versatile forms of SV used in these steps by a unified definition and clarify the essential functionalities that SV can provide for data scientists. We categorize the arts in this field based on the technical challenges they tackled, which include computation efficiency, approximation error, privacy preservation, and appropriate interpretations. We discuss these challenges and analyze the corresponding solutions. We also implement SVBench, the first open-sourced benchmark for developing SV applications, and conduct experiments on six DA tasks to validate our analysis and discussions. Based on the qualitative and quantitative results, we identify the limitations of current efforts for applying SV to DA and highlight the directions of future research and engineering.

A Symmetric Dynamic Learning Framework for Diffeomorphic Medical Image Registration

Diffeomorphic image registration is crucial for various medical imaging applications because it can preserve the topology of the transformation. This study introduces DCCNN-LSTM-Reg, a learning framework that evolves dynamically and learns a symmetrical registration path by satisfying a specified control increment system. This framework aims to obtain symmetric diffeomorphic deformations between moving and fixed images. To achieve this, we combine deep learning networks with diffeomorphic mathematical mechanisms to create a continuous and dynamic registration architecture, which consists of multiple Symmetric Registration (SR) modules cascaded on five different scales. Specifically, our method first uses two U-nets with shared parameters to extract multiscale feature pyramids from the images. We then develop an SR-module comprising a sequential CNN-LSTM architecture to progressively correct the forward and reverse multiscale deformation fields using control increment learning and the homotopy continuation technique. Through extensive experiments on three 3D registration tasks, we demonstrate that our method outperforms existing approaches in both quantitative and qualitative evaluations.

Unsupervised Low-dose CT Reconstruction with One-way Conditional Normalizing Flows

Deep-learning methods have shown promising performance for low-dose computed tomography (LDCT) reconstruction. However, supervised methods face the problem of lacking labeled data in clinical scenarios, and the CNN-based unsupervised denoising methods would cause excessive smoothing in the reconstructed image. Recently, the normalizing flows (NFs) based methods have shown advantages in producing detail-rich images and avoiding over-smoothing, however, there are still issues: (1) Although the alternating optimization in the data and latent space can well utilize the regularization and generation capabilities of NFs, the current two-way transformation strategy of noisy images and latent variables would cause detail loss and secondary artifacts; and (2) Training NFs on high-resolution CT images is hard due to huge computation. Though using conditional normalizing flows (CNFs) to learn conditional probability can reduce the computational burden, current methods require labeled data for conditionalization, and the unsupervised CNFs-based LDCT reconstruction remains a problem. To tackle these problems, we propose a novel CNFs-based unsupervised LDCT iterative reconstruction algorithm. It employs strict one-way transformation when performing alternating optimization in the dual spaces, thus effectively avoiding the problems of detail loss and secondary artifacts. By proposing a novel unsupervised conditionalization strategy, we train CNFs on high-resolution CT images, thus achieving fast and high-quality unsupervised reconstruction. Experiments on different datasets suggest that the performance of the proposed algorithm could surpass some state-of-the-art unsupervised and even supervised methods.

KcMF: A Knowledge-compliant Framework for Schema and Entity Matching with Fine-tuning-free LLMs

Schema and entity matching tasks are crucial for data integration and management. While large language models (LLMs) have shown promising results in these tasks, they suffer from hallucinations and confusion about task instructions. In this paper, we present the Knowledge-Compliant Matching Framework (KcMF), an LLM-based approach that addresses these issues without the need for domain-specific fine-tuning. KcMF employs a pseudo-code-based task decomposition strategy to adopt task-specific natural language statements that guide LLM reasoning and reduce confusion. We also propose two mechanisms, Dataset as Knowledge (DaK) and Example as Knowledge (EaK), to build domain knowledge sets when unstructured domain knowledge is lacking. Additionally, we introduce a result-ensembling strategy to leverage multiple knowledge sources and suppress poorly formatted outputs. Comprehensive evaluations on schema and entity matching tasks demonstrate that KcMF outperforms previous non-LLM state-of-the-art (SOTA) methods by an average F1 score of 22.9% and competes effectively with SOTA fine-tuned LLMs. Moreover, KcMF generalizes well across different LLMs.

Block-to-Scene Pre-training for Point Cloud Hybrid-Domain Masked Autoencoders

Point clouds, as a primary representation of 3D data, can be categorized into scene domain point clouds and object domain point clouds based on the modeled content. Masked autoencoders (MAE) have become the mainstream paradigm in point clouds self-supervised learning. However, existing MAE-based methods are domain-specific, limiting the model's generalization. In this paper, we propose to pre-train a general Point cloud Hybrid-Domain Masked AutoEncoder (PointHDMAE) via a block-to-scene pre-training strategy. We first propose a hybrid-domain masked autoencoder consisting of an encoder and decoder belonging to the scene domain and object domain, respectively. The object domain encoder specializes in handling object point clouds and multiple shared object encoders assist the scene domain encoder in analyzing the scene point clouds. Furthermore, we propose a block-to-scene strategy to pre-train our hybrid-domain model. Specifically, we first randomly select point blocks within a scene and apply a set of transformations to convert each point block coordinates from the scene space to the object space. Then, we employ an object-level mask and reconstruction pipeline to recover the masked points of each block, enabling the object encoder to learn a universal object representation. Finally, we introduce a scene-level block position regression pipeline, which utilizes the blocks' features in the object space to regress these blocks' initial positions within the scene space, facilitating the learning of scene representations. Extensive experiments across different datasets and tasks demonstrate the generalization and superiority of our hybrid-domain model.

Towards Better Generalization: Weight Decay Induces Low-rank Bias for Neural Networks

We study the implicit bias towards low-rank weight matrices when training neural networks (NN) with Weight Decay (WD). We prove that when a ReLU NN is sufficiently trained with Stochastic Gradient Descent (SGD) and WD, its weight matrix is approximately a rank-two matrix. Empirically, we demonstrate that WD is a necessary condition for inducing this low-rank bias across both regression and classification tasks. Our work differs from previous studies as our theoretical analysis does not rely on common assumptions regarding the training data distribution, optimality of weight matrices, or specific training procedures. Furthermore, by leveraging the low-rank bias, we derive improved generalization error bounds and provide numerical evidence showing that better generalization can be achieved. Thus, our work offers both theoretical and empirical insights into the strong generalization performance of SGD when combined with WD.

DS-ViT: Dual-Stream Vision Transformer for Cross-Task Distillation in Alzheimer's Early Diagnosis

In the field of Alzheimer's disease diagnosis, segmentation and classification tasks are inherently interconnected. Sharing knowledge between models for these tasks can significantly improve training efficiency, particularly when training data is scarce. However, traditional knowledge distillation techniques often struggle to bridge the gap between segmentation and classification due to the distinct nature of tasks and different model architectures. To address this challenge, we propose a dual-stream pipeline that facilitates cross-task and cross-architecture knowledge sharing. Our approach introduces a dual-stream embedding module that unifies feature representations from segmentation and classification models, enabling dimensional integration of these features to guide the classification model. We validated our method on multiple 3D datasets for Alzheimer's disease diagnosis, demonstrating significant improvements in classification performance, especially on small datasets. Furthermore, we extended our pipeline with a residual temporal attention mechanism for early diagnosis, utilizing images taken before the atrophy of patients' brain mass. This advancement shows promise in enabling diagnosis approximately six months earlier in mild and asymptomatic stages, offering critical time for intervention.